TEXT runtime·memclrNoHeapPointers<ABIInternal>(SB),NOSPLIT,$0-16
	ADDV	R4, R5, R6

	// if less than 8 bytes, do one byte at a time
	SGTU	$8, R5, R8
	BNE	R8, out

	// do one byte at a time until 8-aligned
	AND	$7, R4, R8
	BEQ	R8, words
	MOVB	R0, (R4)
	ADDV	$1, R4
	JMP	-4(PC)

words:
	// do 8 bytes at a time if there is room
	ADDV	$-7, R6, R5

	PCALIGN	$16
	SGTU	R5, R4, R8
	BEQ	R8, out

		// addr be page-aligned, and allows length to be zero.
		throw("unaligned sysNoHugePageOS")
	}
	madvise(v, n, _MADV_NOHUGEPAGE)
}

func sysHugePageCollapseOS(v unsafe.Pointer, n uintptr) {
	if uintptr(v)&(physPageSize-1) != 0 {
		// The Linux implementation requires that the address
		// addr be page-aligned, and allows length to be zero.
		throw("unaligned sysHugePageCollapseOS")
	}

	// nil pointer is always suitably aligned (#47430).
	if p == nil {
		return
	}

	// Check that (*[n]elem)(p) is appropriately aligned.
	// Note that we allow unaligned pointers if the types they point to contain
	// no pointers themselves. See issue 37298.
	// TODO(mdempsky): What about fieldAlign?
	if elem.Pointers() && uintptr(p)&(uintptr(elem.Align_)-1) != 0 {
		throw("checkptr: misaligned pointer conversion")

entry:
	ADDV	R4, R14, R12	// R6 start of a, R14 end of a
	BEQ	R4, R12, samebytes // length is 0

	SRLV	$4, R14		// R14 is number of chunks
	BEQ	R0, R14, byte_loop

	// make sure both a and b are aligned.
	OR	R4, R6, R15
	AND	$7, R15
	BNE	R0, R15, byte_loop

	PCALIGN	$16
chunk16_loop:
	BEQ	R0, R14, byte_loop
	MOVV	(R4), R8
	MOVV	(R6), R9
	BNE	R8, R9, byte_loop
	MOVV	8(R4), R16
	MOVV	8(R6), R17

    TF_DataType, const int64_t* dims, int num_dims, void* data, size_t len,
    void (*deallocator)(void* data, size_t len, void* arg),
    void* deallocator_arg);

// Returns the alignment, in bytes, required for allocating aligned tensors.
//
// This can be used in combination with TF_NewTensor to manually manage
// memory while ensuring the resulting tensors satisfy TensorFlow's
// memory alignment preferences.

TEXT ·Or8(SB),NOSPLIT,$0-5
	MOVW	ptr+0(FP), R1
	MOVBU	val+4(FP), R2
	MOVW	$~3, R3	// Align ptr down to 4 bytes so we can use 32-bit load/store.
	AND	R1, R3
#ifdef GOARCH_mips
	// Big endian.  ptr = ptr ^ 3
	XOR	$3, R1
#endif
	AND	$3, R1, R4	// R4 = ((ptr & 3) * 8)
	SLL	$3, R4
	SLL	R4, R2, R2	// Shift val for aligned ptr. R2 = val << R4
	SYNC
try_or8:
	LL	(R3), R4	// R4 = *R3
	OR	R2, R4

 * units away from 0 and 0.499 towards 0. The smallest unit this returns is "µs"; the largest unit
 * it returns is "s". For values in [-499..499] this returns "  0 µs".
 *
 * The returned string attempts to be column-aligned to 6 characters. For negative and large values
 * the returned string may be longer.
 */
fun formatDuration(ns: Long): String {
  val s =
    when {
      ns <= -999_500_000 -> "${(ns - 500_000_000) / 1_000_000_000} s "

// Lock is a no-op used by -copylocks checker from `go vet`.
func (*noCopy) Lock()   {}
func (*noCopy) Unlock() {}

// align64 may be added to structs that must be 64-bit aligned.
// This struct is recognized by a special case in the compiler
// and will not work if copied to any other package.

		for i, u := range util {
			if u.Time+int64(window) > util[len(util)-1].Time {
				break
			}
			mmu = math.Min(mmu, muInWindow(util[i:], u.Time+int64(window)))
		}
	}

	// Consider all left-aligned windows.
	update()
	// Reverse the trace. Slightly subtle because each MutatorUtil
	// is a *change*.
	rutil := make([]trace.MutatorUtil, len(util))
	if util[len(util)-1].Util != 0 {
		panic("irreversible trace")

// available in the pool.
func (bp *BytePoolCap) Get() (b []byte) {
	if bp == nil {
		return nil
	}
	select {
	case b = <-bp.c:
		// reuse existing buffer
	default:
		// create new aligned buffer
		if bp.wcap > 0 {
			b = reedsolomon.AllocAligned(1, bp.wcap)[0][:bp.w]
		} else {
			b = reedsolomon.AllocAligned(1, bp.w)[0]
		}
	}
	return
}

// Put returns the given Buffer to the BytePool.